- raw http://vserver.13thfloor.at/Experimental/patch-2.6.10-vs1.9.3.17.diff

[packages/kernel.git] / linux-cluster-gnbd.patch

diff -urN linux-2.6.9/drivers/block/Kconfig linux-2.6.9-patched/drivers/block/Kconfig
--- linux-2.6.9/drivers/block/Kconfig	2004-10-18 16:53:43.000000000 -0500
+++ linux-2.6.9-patched/drivers/block/Kconfig	2004-10-22 13:27:52.938836304 -0500
@@ -356,6 +356,13 @@
 	  your machine, or if you want to have a raid or loopback device
 	  bigger than 2TB.  Otherwise say N.
 
+config BLK_DEV_GNBD
+	tristate "Global network block device support"
+	depends on NET
+	---help---
+	  
+	  If unsure, say N.
+
 source "drivers/s390/block/Kconfig"
 
 endmenu
diff -urN linux-2.6.9/drivers/block/Makefile linux-2.6.9-patched/drivers/block/Makefile
--- linux-2.6.9/drivers/block/Makefile	2004-10-18 16:54:55.000000000 -0500
+++ linux-2.6.9-patched/drivers/block/Makefile	2004-10-22 13:30:48.224188880 -0500
@@ -43,4 +43,4 @@
 obj-$(CONFIG_VIODASD)		+= viodasd.o
 obj-$(CONFIG_BLK_DEV_SX8)	+= sx8.o
 obj-$(CONFIG_BLK_DEV_UB)	+= ub.o
-
+obj-$(CONFIG_BLK_DEV_GNBD)	+= gnbd.o
diff -urN linux-2.6.9/drivers/block/gnbd.c linux-2.6.9-patched/drivers/block/gnbd.c
--- linux-2.6.9/drivers/block/gnbd.c	1969-12-31 18:00:00.000000000 -0600
+++ linux-2.6.9-patched/drivers/block/gnbd.c	2004-10-22 13:43:33.303879088 -0500
@@ -0,0 +1,1053 @@
+/******************************************************************************
+*******************************************************************************
+**
+**  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
+**  Copyright (C) 2004 Red Hat, Inc.  All rights reserved.
+**
+**  This copyrighted material is made available to anyone wishing to use,
+**  modify, copy, or redistribute it subject to the terms and conditions
+**  of the GNU General Public License v.2.
+**
+*******************************************************************************
+******************************************************************************/
+
+/* Large chunks of this code were lifted from nbd.c */
+
+#include <linux/major.h>
+
+#include <linux/blkdev.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/fs.h>
+#include <linux/bio.h>
+#include <linux/stat.h>
+#include <linux/errno.h>
+#include <linux/file.h>
+#include <linux/ioctl.h>
+#include <net/sock.h>
+#include <linux/in.h>
+#include <linux/buffer_head.h>
+#include <linux/miscdevice.h>
+
+#include <linux/devfs_fs_kernel.h>
+
+#include <asm/uaccess.h>
+#include <asm/types.h>
+
+#include <linux/gnbd.h>
+
+static int major_nr = 0;
+uint64_t insmod_time;
+
+
+#define GNBD_MAGIC 0x74d06100
+
+#ifdef NDEBUG
+#define dprintk(flags, fmt...)
+#else /* NDEBUG */
+#define dprintk(flags, fmt...) do { \
+	if (debugflags & (flags)) printk(KERN_DEBUG fmt); \
+} while (0)
+#define DBG_IOCTL       0x0004
+#define DBG_INIT        0x0010
+#define DBG_EXIT        0x0020
+#define DBG_BLKDEV      0x0100
+#define DBG_RX          0x0200
+#define DBG_TX          0x0400
+static unsigned int debugflags;
+#endif /* NDEBUG */
+
+static struct gnbd_device gnbd_dev[MAX_GNBD];
+
+struct request shutdown_req;
+struct request ping_req;
+
+static spinlock_t gnbd_lock = SPIN_LOCK_UNLOCKED;
+
+#define to_gnbd_dev(d) container_of(d, struct gnbd_device, class_dev)
+
+static void gnbd_class_release(struct class_device *class_dev)
+{
+	printk("releasing gnbd class\n");
+	/* FIXME -- What the hell do I have to free up here */
+}
+
+static struct class gnbd_class = {
+	.name = "gnbd",
+	.release = gnbd_class_release
+};
+
+
+static ssize_t show_pid(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	return sprintf(buf, "%d\n", dev->receiver_pid);
+}
+
+static CLASS_DEVICE_ATTR(pid, S_IRUGO, show_pid, NULL);
+
+static ssize_t show_server(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	if (dev->server_name)
+		return sprintf(buf, "%s/%hx\n", dev->server_name,
+				dev->server_port);
+	else
+		return sprintf(buf, "\n");
+}
+
+/* FIXME -- should a empty store free the memory */
+static ssize_t store_server(struct class_device *class_dev,
+		const char *buf, size_t count)
+{
+	int res;
+	short unsigned int port;
+	char *ptr;
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	if (down_trylock(&dev->do_it_lock))
+		return -EBUSY;
+	if (dev->server_name)
+		kfree(dev->server_name);
+	dev->server_name = kmalloc(count + 1, GFP_KERNEL);
+	if (!dev->server_name)
+		return -ENOMEM;
+	memcpy(dev->server_name, buf, count);
+	dev->server_name[count] = 0;
+	ptr = strchr(dev->server_name, '/');
+	if (!ptr)
+		return -EINVAL;
+	*ptr++ = 0;
+	res = sscanf(ptr, "%4hx", &port);
+	if (res != 1){
+		up(&dev->do_it_lock);
+		return -EINVAL;
+	}
+	dev->server_port = port;
+	up(&dev->do_it_lock);
+	return count;
+}
+
+CLASS_DEVICE_ATTR(server, S_IRUGO | S_IWUSR, show_server, store_server);
+
+static ssize_t show_name(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	return sprintf(buf, "%s\n", dev->name);
+}
+
+static ssize_t store_name(struct class_device *class_dev,
+                const char *buf, size_t count)
+{
+	int res;
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	if (down_trylock(&dev->do_it_lock))
+		return -EBUSY;
+	res = sscanf(buf, "%31s", dev->name);
+	up(&dev->do_it_lock);
+	if (res != 1)
+		return -EINVAL;
+	return count;
+}
+
+CLASS_DEVICE_ATTR(name, S_IRUGO | S_IWUSR, show_name, store_name);
+
+
+static ssize_t show_sectors(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	return sprintf(buf, "%Lu\n",
+			(unsigned long long)get_capacity(dev->disk));
+}
+
+static ssize_t store_sectors(struct class_device *class_dev,
+		const char *buf, size_t count)
+{
+	int res;
+	sector_t size;
+	struct block_device *bdev;
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	
+	if (down_trylock(&dev->do_it_lock))
+		return -EBUSY;
+	res = sscanf(buf, "%Lu\n", &size);
+	if (res != 1){
+		up(&dev->do_it_lock);
+		return -EINVAL;
+	}
+	/* FIXME -- should I switch the order here, so that I don't have
+	   capacity set to one thing and the bdev inode size set to another */ 
+	set_capacity(dev->disk, size);
+	bdev = bdget_disk(dev->disk, 0);
+	if (bdev) {
+		down(&bdev->bd_inode->i_sem);
+		i_size_write(bdev->bd_inode, (loff_t)size << 9);
+		up(&bdev->bd_inode->i_sem);
+		bdput(bdev);
+	}
+	up(&dev->do_it_lock);
+	return count;
+}
+
+CLASS_DEVICE_ATTR(sectors, S_IRUGO | S_IWUSR, show_sectors, store_sectors);
+
+static ssize_t show_usage(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	return sprintf(buf, "%d\n", dev->open_count);
+}
+
+CLASS_DEVICE_ATTR(usage, S_IRUGO, show_usage, NULL);
+
+static ssize_t show_flags(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	return sprintf(buf, "0x%04x\n", dev->flags);
+}
+
+static ssize_t store_flags(struct class_device *class_dev,
+                const char *buf, size_t count)
+{
+	int res;
+	
+        struct gnbd_device *dev = to_gnbd_dev(class_dev);
+        if (down_trylock(&dev->do_it_lock))
+                return -EBUSY;
+	res = sscanf(buf, "0x%hx", &dev->flags);
+	up(&dev->do_it_lock);
+        if (res != 1)
+                return -EINVAL;
+        return count;
+}
+
+
+CLASS_DEVICE_ATTR(flags, S_IRUGO | S_IWUSR, show_flags, store_flags);
+
+static ssize_t show_waittime(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	if (list_empty(&dev->queue_head))
+		return sprintf(buf, "-1\n");
+	return sprintf(buf, "%ld\n",
+			((long)jiffies - (long)dev->last_received) / HZ);
+}
+
+CLASS_DEVICE_ATTR(waittime, S_IRUGO, show_waittime, NULL);
+
+static ssize_t show_connected(struct class_device *class_dev, char *buf)
+{
+	struct gnbd_device *dev = to_gnbd_dev(class_dev);
+	return sprintf(buf, "%d\n", (dev->sock != NULL));
+}
+
+CLASS_DEVICE_ATTR(connected, S_IRUGO, show_connected, NULL);
+
+#ifndef NDEBUG
+static const char *ioctl_cmd_to_ascii(int cmd)
+{
+	switch (cmd) {
+	case GNBD_DO_IT: return "do-it";
+	case GNBD_CLEAR_QUE: return "clear-que";
+	case GNBD_PRINT_DEBUG: return "print-debug";
+	case GNBD_DISCONNECT: return "disconnect";
+	}
+	return "unknown";
+}
+
+static const char *gnbdcmd_to_ascii(int cmd)
+{
+	switch (cmd) {
+	case  GNBD_CMD_READ: return "read";
+	case GNBD_CMD_WRITE: return "write";
+	case  GNBD_CMD_DISC: return "disconnect";
+	case GNBD_CMD_PING: return "ping";
+	}
+	return "invalid";
+}
+#endif /* NDEBUG */
+
+static void gnbd_end_request(struct request *req)
+{
+	int uptodate = (req->errors == 0) ? 1 : 0;
+	request_queue_t *q = req->q;
+	unsigned long flags;
+
+	dprintk(DBG_BLKDEV, "%s: request %p: %s\n", req->rq_disk->disk_name,
+			req, uptodate? "done": "failed");
+
+	if (!uptodate)
+		printk("%s %d called gnbd_end_request with and error\n",
+		       current->comm, current->pid);	
+	
+	spin_lock_irqsave(q->queue_lock, flags);
+	if (!end_that_request_first(req, uptodate, req->nr_sectors)) {
+		end_that_request_last(req);
+	}
+	spin_unlock_irqrestore(q->queue_lock, flags);
+}
+
+/*
+ *  Send or receive packet.
+ */
+static int sock_xmit(struct socket *sock, int send, void *buf, int size,
+		int msg_flags)
+{
+	mm_segment_t oldfs;
+	int result;
+	struct msghdr msg;
+	struct iovec iov;
+	unsigned long flags;
+	sigset_t oldset;
+
+	oldfs = get_fs();
+	set_fs(get_ds());
+	/* Allow interception of SIGKILL only
+	 * Don't allow other signals to interrupt the transmission */
+	spin_lock_irqsave(&current->sighand->siglock, flags);
+	oldset = current->blocked;
+	sigfillset(&current->blocked);
+	sigdelsetmask(&current->blocked, sigmask(SIGKILL) | sigmask(SIGTERM) |
+	              sigmask(SIGHUP));
+	recalc_sigpending();
+	spin_unlock_irqrestore(&current->sighand->siglock, flags);
+
+	do {
+		sock->sk->sk_allocation = GFP_NOIO;
+		iov.iov_base = buf;
+		iov.iov_len = size;
+		msg.msg_name = NULL;
+		msg.msg_namelen = 0;
+		msg.msg_iov = &iov;
+		msg.msg_iovlen = 1;
+		msg.msg_control = NULL;
+		msg.msg_controllen = 0;
+		msg.msg_namelen = 0;
+		msg.msg_flags = msg_flags | MSG_NOSIGNAL;
+
+		if (send)
+			result = sock_sendmsg(sock, &msg, size);
+		else
+			result = sock_recvmsg(sock, &msg, size, 0);
+
+		if (signal_pending(current)) {
+			siginfo_t info;
+			spin_lock_irqsave(&current->sighand->siglock, flags);
+			printk(KERN_WARNING "gnbd (pid %d: %s) got signal %d\n",
+				current->pid, current->comm, 
+				dequeue_signal(current, &current->blocked, &info));
+			spin_unlock_irqrestore(&current->sighand->siglock, flags);
+			result = -EINTR;
+			break;
+		}
+
+		if (result <= 0) {
+			if (result == 0)
+				result = -EPIPE; /* short read */
+			break;
+		}
+		size -= result;
+		buf += result;
+	} while (size > 0);
+
+	spin_lock_irqsave(&current->sighand->siglock, flags);
+	current->blocked = oldset;
+	recalc_sigpending();
+	spin_unlock_irqrestore(&current->sighand->siglock, flags);
+
+	set_fs(oldfs);
+	return result;
+}
+
+static inline int sock_send_bvec(struct socket *sock, struct bio_vec *bvec,
+		int flags)
+{
+	int result;
+	void *kaddr = kmap(bvec->bv_page);
+	result = sock_xmit(sock, 1, kaddr + bvec->bv_offset, bvec->bv_len,
+			flags);
+	kunmap(bvec->bv_page);
+	return result;
+}
+
+
+#define gnbd_send_req(dev, req) __gnbd_send_req((dev), (dev)->sock, (req))
+	
+int __gnbd_send_req(struct gnbd_device *dev, struct socket *sock,
+		struct request *req)
+{
+	int result, i, flags;
+	struct gnbd_request request;
+	unsigned long size = req->nr_sectors << 9;
+
+	request.magic = htonl(GNBD_REQUEST_MAGIC);
+	request.type = htonl(gnbd_cmd(req));
+	request.from = cpu_to_be64((u64) req->sector << 9);
+	request.len = htonl(size);
+	memcpy(request.handle, &req, sizeof(req));
+
+	down(&dev->tx_lock);
+
+	if (!sock) {
+		printk(KERN_ERR "%s: Attempted send on closed socket\n",
+				dev->disk->disk_name);
+		result = -ENOTCONN;
+		goto error_out;
+	}
+
+	dprintk(DBG_TX, "%s: request %p: sending control (%s@%llu,%luB)\n",
+			dev->disk->disk_name, req,
+			gnbdcmd_to_ascii(gnbd_cmd(req)),
+			(unsigned long long)req->sector << 9,
+			req->nr_sectors << 9);
+	result = sock_xmit(sock, 1, &request, sizeof(request),
+			(gnbd_cmd(req) == GNBD_CMD_WRITE)? MSG_MORE: 0);
+	if (result < 0) {
+		printk(KERN_ERR "%s: Send control failed (result %d)\n",
+				dev->disk->disk_name, result);
+		goto error_out;
+	}
+
+	if (gnbd_cmd(req) == GNBD_CMD_WRITE) {
+		struct bio *bio;
+		/*
+		 * we are really probing at internals to determine
+		 * whether to set MSG_MORE or not...
+		 */
+		rq_for_each_bio(bio, req) {
+			struct bio_vec *bvec;
+			bio_for_each_segment(bvec, bio, i) {
+				flags = 0;
+				if ((i < (bio->bi_vcnt - 1)) || bio->bi_next)
+					flags = MSG_MORE;
+				dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n",
+						dev->disk->disk_name, req,
+						bvec->bv_len);
+				result = sock_send_bvec(sock, bvec, flags);
+				if (result < 0) {
+					printk(KERN_ERR "%s: Send data failed (result %d)\n",
+							dev->disk->disk_name,
+							result);
+					goto error_out;
+				}
+			}
+		}
+	}
+	up(&dev->tx_lock);
+	return 0;
+
+error_out:
+	up(&dev->tx_lock);
+	return result;
+}
+
+	
+static int gnbd_find_request(struct gnbd_device *dev, struct request *xreq)
+{
+	struct request *req;
+	struct list_head *tmp;
+
+	list_for_each(tmp, &dev->queue_head) {
+		req = list_entry(tmp, struct request, queuelist);
+		if (req != xreq)
+			continue;
+		return 1;
+	}
+	return 0;
+}
+
+static inline int sock_recv_bvec(struct socket *sock, struct bio_vec *bvec)
+{
+	int result;
+	void *kaddr = kmap(bvec->bv_page);
+	result = sock_xmit(sock, 0, kaddr + bvec->bv_offset, bvec->bv_len,
+			MSG_WAITALL);
+	kunmap(bvec->bv_page);
+	return result;
+}
+
+int gnbd_recv_req(struct gnbd_device *dev, struct request *req)
+{
+	int result;
+	int i;
+	struct bio *bio;
+	rq_for_each_bio(bio, req) {
+		struct bio_vec *bvec;
+		bio_for_each_segment(bvec, bio, i) {
+			result = sock_recv_bvec(dev->sock, bvec);
+			if (result < 0) {
+				printk(KERN_ERR "%s: Receive data failed (result %d)\n",
+						dev->disk->disk_name,
+						result);
+				return result;
+			}
+			dprintk(DBG_RX, "%s: request %p: got %d bytes data\n",
+					dev->disk->disk_name, req, bvec->bv_len);
+		}
+	}
+	return 0;
+}
+
+int gnbd_do_it(struct gnbd_device *dev)
+{
+	int result;
+	struct gnbd_reply reply;
+	struct request *req;
+	struct socket *sock = dev->sock;
+
+	BUG_ON(dev->magic != GNBD_MAGIC);
+
+	while((result = sock_xmit(sock, 0, &reply,sizeof(reply), MSG_WAITALL)) > 0){
+		if (ntohl(reply.magic) == GNBD_KEEP_ALIVE_MAGIC)
+			/* FIXME -- I should reset the wait time here */
+			continue;
+
+		memcpy(&req, reply.handle, sizeof(req));
+		if (req == &shutdown_req)
+			return 0;
+
+		if (!gnbd_find_request(dev, req)){
+			printk(KERN_ERR "%s: Unexpected reply (%p)\n",
+					dev->disk->disk_name, reply.handle);
+			return -EBADR;
+		}
+		if (ntohl(reply.magic) != GNBD_REPLY_MAGIC) {
+			printk(KERN_ERR "%s: Wrong magic (0x%lx)\n",
+					dev->disk->disk_name,
+					(unsigned long)ntohl(reply.magic));
+			return -EPROTO;
+		}
+		if (ntohl(reply.error)) {
+			printk(KERN_ERR "%s: Other side returned error (%d)\n",
+					dev->disk->disk_name, ntohl(reply.error));
+			req->errors++;
+			goto remove_req;
+		}
+		dprintk(DBG_RX, "%s: request %p: got reply\n",
+				dev->disk->disk_name, req);
+
+		if (gnbd_cmd(req) == GNBD_CMD_READ){
+			result = gnbd_recv_req(dev, req);
+			if (result < 0)
+				return result;
+		}
+remove_req:
+		spin_lock(&dev->queue_lock);
+		list_del_init(&req->queuelist);
+		dev->last_received = jiffies;
+		spin_unlock(&dev->queue_lock);
+		if (req != &ping_req)
+			gnbd_end_request(req);
+	}
+	printk(KERN_ERR "%s: Receive control failed (result %d)\n",
+			dev->disk->disk_name, result);
+	return result;
+}
+
+void gnbd_clear_que(struct gnbd_device *dev)
+{
+	struct request *req;
+
+	BUG_ON(dev->magic != GNBD_MAGIC);
+
+	do {
+		req = NULL;
+		if (!list_empty(&dev->queue_head)) {
+			req = list_entry(dev->queue_head.next, struct request, queuelist);
+			list_del_init(&req->queuelist);
+		}
+		if (req && req != &ping_req) {
+			req->errors++;
+			gnbd_end_request(req);
+		}
+	} while (req);
+}
+
+/*
+ * We always wait for result of write, for now. It would be nice to make it optional
+ * in future
+ * if ((req->cmd == WRITE) && (dev->flags & GNBD_WRITE_NOCHK)) 
+ *   { printk( "Warning: Ignoring result!\n"); gnbd_end_request( req ); }
+ */
+
+static void do_gnbd_request(request_queue_t * q)
+{
+	int err;
+	struct request *req;
+	
+	while ((req = elv_next_request(q)) != NULL) {
+		struct gnbd_device *dev;
+
+		blkdev_dequeue_request(req);
+		dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%lx)\n",
+				req->rq_disk->disk_name, req, req->flags);
+
+		if (!(req->flags & REQ_CMD))
+			goto error_out;
+		
+		dev = req->rq_disk->private_data;
+
+		if (dev->receiver_pid == -1)
+			goto error_out;
+		
+		BUG_ON(dev->magic != GNBD_MAGIC);
+
+		gnbd_cmd(req) = GNBD_CMD_READ;
+		if (rq_data_dir(req) == WRITE) {
+			gnbd_cmd(req) = GNBD_CMD_WRITE;
+			if (dev->flags & GNBD_READ_ONLY) {
+				printk(KERN_ERR "%s: Write on read-only\n",
+						dev->disk->disk_name);
+				goto error_out;
+			}
+		}
+
+		req->errors = 0;
+		spin_unlock_irq(q->queue_lock);
+
+		spin_lock(&dev->queue_lock);
+
+		if (list_empty(&dev->queue_head))
+			dev->last_received = jiffies;
+		list_add(&req->queuelist, &dev->queue_head);
+		spin_unlock(&dev->queue_lock);
+
+		err = gnbd_send_req(dev, req);
+
+		spin_lock_irq(q->queue_lock);
+		if (err)
+			goto sock_error;
+		continue;
+
+error_out:
+		req->errors++;
+		spin_unlock(q->queue_lock);
+		gnbd_end_request(req);
+		spin_lock(q->queue_lock);
+	}
+	return;
+
+sock_error:
+	return;
+}
+
+/*
+ * This is called before dev-sock is set, so you dodn't need
+ * to worry about the tx_lock or the queue_lock
+ */
+static int gnbd_resend_requests(struct gnbd_device *dev, struct socket *sock)
+{
+	int err = 0;
+	struct request *req;
+	struct list_head *tmp;
+	
+	printk("resending requests\n");
+	list_for_each(tmp, &dev->queue_head) {
+		req = list_entry(tmp, struct request, queuelist);
+		err = __gnbd_send_req(dev, sock, req);
+
+		if (err){
+			printk("failed trying to resend request (%d)\n", err);
+			break;
+		}
+	}
+
+	return err;
+}
+/*
+static int get_server_info(struct gnbd_device *dev, struct socket *sock)
+{
+	struct sockaddr_in server;
+	int len;
+	int err;
+
+	err = sock->ops->getname(sock, (struct sockaddr *) &server, &len, 1);
+	if (err) {
+		printk(KERN_WARNING "cannot get socket info, shutting down\n");
+	} else{
+		dev->server_addr = server.sin_addr;
+		dev->server_port = server.sin_port;
+	}
+	return err;
+}
+*/
+
+static int gnbd_ctl_ioctl(struct inode *inode, struct file *file,
+		     unsigned int cmd, unsigned long arg)
+{
+	struct gnbd_device *dev = NULL;
+	struct block_device *bdev;
+        do_it_req_t req;
+	int error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (cmd == GNBD_DISCONNECT || cmd == GNBD_CLEAR_QUE ||
+                        cmd == GNBD_PING || cmd == GNBD_PRINT_DEBUG) {
+                if (arg >= MAX_GNBD)
+                        return -EINVAL;
+                dev = &gnbd_dev[arg];
+                BUG_ON(dev->magic != GNBD_MAGIC);
+        }
+
+	/* Anyone capable of this syscall can do *real bad* things */
+	dprintk(DBG_IOCTL, "%s: gnbd_ioctl cmd=%s(0x%x) arg=%lu\n",
+			dev->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg);
+
+	switch (cmd) {
+	case GNBD_DISCONNECT:
+	        printk(KERN_INFO "%s: GNBD_DISCONNECT\n", dev->disk->disk_name);
+		spin_lock(&dev->open_lock);
+		if (dev->open_count > 0){
+			spin_unlock(&dev->open_lock);
+			return -EBUSY;
+		}
+		dev->receiver_pid = -1;
+		spin_unlock(&dev->open_lock);
+		/* There is no one using the device, you can disconnect it */
+		if (dev->sock == NULL)
+			return -ENOTCONN;
+		gnbd_send_req(dev, &shutdown_req);
+                return 0;
+	case GNBD_CLEAR_QUE:
+		if (down_interruptible(&dev->do_it_lock))
+			return -EBUSY;
+		dev->receiver_pid = -1;
+		gnbd_clear_que(dev);
+		bdev = dev->bdev;
+		if (bdev) {
+			blk_run_queue(dev->disk->queue);
+			fsync_bdev(bdev);
+			invalidate_bdev(bdev, 0);
+		}
+		up(&dev->do_it_lock);
+		return 0;
+	case GNBD_DO_IT:
+		if (copy_from_user(&req, (do_it_req_t *)arg, sizeof(req)))
+                        return -EFAULT;
+                if (req.minor >= 128)
+                        return -EINVAL;
+                dev = &gnbd_dev[req.minor];
+                BUG_ON(dev->magic != GNBD_MAGIC);
+		if (dev->file)
+			return -EBUSY;
+		error = -EINVAL;
+		file = fget(req.sock_fd);
+		if (!file)
+			return error;
+		inode = file->f_dentry->d_inode;
+		if (!inode->i_sock) {
+			fput(file);
+			return error;
+		}
+		if (down_trylock(&dev->do_it_lock)){
+			fput(file);
+			return -EBUSY;
+		}
+		error = gnbd_resend_requests(dev, SOCKET_I(inode));
+		if (error){
+			printk("quitting NBD_DO_IT\n");
+			up(&dev->do_it_lock);
+			fput(file);
+			return error;
+		}
+		dev->file = file;
+		dev->sock = SOCKET_I(inode);
+		dev->receiver_pid = current->pid; 
+		blk_run_queue(dev->disk->queue);
+		error = gnbd_do_it(dev);
+		/* should I kill the socket first */
+		up(&dev->do_it_lock);
+		down(&dev->tx_lock);
+		if (dev->sock) {
+			printk(KERN_WARNING "%s: shutting down socket\n",
+					dev->disk->disk_name);
+			dev->sock->ops->shutdown(dev->sock,
+					SEND_SHUTDOWN|RCV_SHUTDOWN);
+			dev->sock = NULL;
+		}
+		up(&dev->tx_lock);
+		file = dev->file;
+		dev->file = NULL;
+		if (file)
+			fput(file);
+		printk("exitting GNBD_DO_IT ioctl\n");
+		return error;
+	case GNBD_PING:
+		/* FIXME -- should I allow pings if everything is compeletely
+		 * shutdown */
+		spin_lock(&dev->queue_lock);
+		/* only one outstanding ping at a time */
+		if (list_empty(&ping_req.queuelist)){
+			if (list_empty(&dev->queue_head))
+				dev->last_received = jiffies;
+			list_add(&ping_req.queuelist, &dev->queue_head);
+		}
+		spin_unlock(&dev->queue_lock);
+		gnbd_send_req(dev, &ping_req); /* ignore the errors */
+		return 0;
+	case GNBD_PRINT_DEBUG:
+		printk(KERN_INFO "%s: next = %p, prev = %p, head = %p\n",
+			dev->disk->disk_name,
+			dev->queue_head.next, dev->queue_head.prev,
+			&dev->queue_head);
+		return 0;
+	case GNBD_GET_TIME:
+		if (copy_to_user((void *)arg, &insmod_time, sizeof(uint64_t))){
+			printk(KERN_WARNING "couldn't compy time argument to user\n");
+			return -EFAULT;
+		}
+		return 0;
+	}
+	/* FIXME -- should I print something, is EINVAL the right error */
+	return -EINVAL;
+}
+
+static int gnbd_open(struct inode *inode, struct file *file)
+{
+	struct gnbd_device *dev = inode->i_bdev->bd_disk->private_data;
+	spin_lock(&dev->open_lock);
+	if (dev->receiver_pid == -1){
+		spin_unlock(&dev->open_lock);
+		return -ENXIO;
+	}
+	spin_unlock(&dev->open_lock);
+	if ((file->f_mode & FMODE_WRITE) && (dev->flags & GNBD_READ_ONLY)){
+		printk(KERN_INFO "cannot open read only gnbd device read/write");
+		return -EROFS;
+	}
+
+	dev->open_count++;
+	dev->bdev = inode->i_bdev;
+	return 0;
+}
+
+/* FIXME -- I don't sync the device at close. This means that If you write
+ * something, and close the device, and expect that then it is written,
+ * you are wrong.... This might cause problems */
+static int gnbd_release(struct inode *inode, struct file *file)
+{
+	struct gnbd_device *dev = inode->i_bdev->bd_disk->private_data;
+
+	dev->open_count--;
+	if (dev->open_count == 0)
+		dev->bdev = NULL;
+	return 0;
+}
+
+static struct file_operations _gnbd_ctl_fops =
+{
+        .ioctl = gnbd_ctl_ioctl,
+        .owner = THIS_MODULE,
+};
+
+static struct miscdevice _gnbd_misc =
+{
+        .minor = MISC_DYNAMIC_MINOR,
+        .name  = "gnbd_ctl",
+        .devfs_name = "gnbd_ctl",
+        .fops = &_gnbd_ctl_fops
+};
+
+/* FIXME -- I should probably do more here */
+int __init gnbd_ctl_init(void)
+{
+        int err;
+        
+        err = misc_register(&_gnbd_misc);
+        if (err) {
+                printk("cannot register control device\n");
+                return err;
+        }
+        return 0;
+}
+
+void gnbd_ctl_cleanup(void)
+{
+        if (misc_deregister(&_gnbd_misc) < 0)
+                printk("cannot deregister control device\n");
+}
+
+static struct block_device_operations gnbd_fops =
+{
+	.open =		gnbd_open,
+	.release =	gnbd_release,
+	.owner =	THIS_MODULE,
+};
+
+/*
+ * And here should be modules and kernel interface 
+ *  (Just smiley confuses emacs :-)
+ */
+
+static int __init gnbd_init(void)
+{
+	int err = -ENOMEM;
+	struct timeval tv;
+	int i;
+
+	if (sizeof(struct gnbd_request) != 28) {
+		printk(KERN_CRIT "gnbd: sizeof gnbd_request needs to be 28 in order to work!\n" );
+		return -EIO;
+	}
+	shutdown_req.flags = REQ_SPECIAL;
+	gnbd_cmd(&shutdown_req) = GNBD_CMD_DISC;
+	shutdown_req.sector = 0;
+	shutdown_req.nr_sectors = 0;
+
+	ping_req.flags = REQ_SPECIAL;
+	gnbd_cmd(&ping_req) = GNBD_CMD_PING;
+	ping_req.sector = 0;
+	ping_req.nr_sectors = 0;
+	
+	for (i = 0; i < MAX_GNBD; i++) {
+		struct gendisk *disk = alloc_disk(1);
+		if (!disk)
+			goto out;
+		gnbd_dev[i].disk = disk;
+		/*
+		 * The new linux 2.5 block layer implementation requires
+		 * every gendisk to have its very own request_queue struct.
+		 * These structs are big so we dynamically allocate them.
+		 */
+		disk->queue = blk_init_queue(do_gnbd_request, &gnbd_lock);
+		if (!disk->queue) {
+			put_disk(disk);
+			goto out;
+		}
+	}
+	major_nr = register_blkdev(major_nr, "gnbd");
+	if (major_nr < 0) {
+		printk("gnbd: unable to get a major number\n");
+		err = major_nr;
+		goto out;
+	}
+
+	printk(KERN_INFO "gnbd: registered device at major %d\n", major_nr);
+	dprintk(DBG_INIT, "gnbd: debugflags=0x%x\n", debugflags);
+
+	devfs_mk_dir("gnbd_minor");
+	err = class_register(&gnbd_class);
+	if (err)
+		goto out_unregister;
+	for (i = 0; i < MAX_GNBD; i++) {
+		struct gendisk *disk = gnbd_dev[i].disk;
+		gnbd_dev[i].file = NULL;
+		gnbd_dev[i].magic = GNBD_MAGIC;
+		gnbd_dev[i].flags = 0;
+		gnbd_dev[i].open_count = 0;
+		gnbd_dev[i].receiver_pid = -1;
+		gnbd_dev[i].server_name = NULL;
+		gnbd_dev[i].server_port = 0;
+		gnbd_dev[i].name[0] = '\0';
+		gnbd_dev[i].bdev = NULL;
+		spin_lock_init(&gnbd_dev[i].queue_lock);
+		spin_lock_init(&gnbd_dev[i].open_lock);
+		INIT_LIST_HEAD(&gnbd_dev[i].queue_head);
+		init_MUTEX(&gnbd_dev[i].tx_lock);
+		init_MUTEX(&gnbd_dev[i].do_it_lock);
+		gnbd_dev[i].class_dev.class = &gnbd_class;
+		sprintf(gnbd_dev[i].class_dev.class_id, "gnbd%d", i);
+		err = class_device_register(&gnbd_dev[i].class_dev);
+		if (err){
+			printk("class_device_register failed with %d\n", err);
+			goto out_unregister_class;
+		}
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_pid))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_server))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_name))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_sectors))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_usage))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_flags))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_waittime))
+			goto out_remove_file;
+		if(class_device_create_file(&gnbd_dev[i].class_dev,
+					&class_device_attr_connected))
+			goto out_remove_file;
+		disk->major = major_nr;
+		disk->first_minor = i;
+		disk->fops = &gnbd_fops;
+		disk->private_data = &gnbd_dev[i];
+		sprintf(disk->disk_name, "gnbd%d", i);
+		sprintf(disk->devfs_name, "gnbd_minor/%d", i);
+		set_capacity(disk, 0);
+		add_disk(disk);
+		if(sysfs_create_link(&gnbd_dev[i].class_dev.kobj,
+					&gnbd_dev[i].disk->kobj, "block"))
+			goto out_remove_disk;
+		
+	}
+
+        err = gnbd_ctl_init();
+        if (err)
+                goto out_unregister_class;
+        
+	insmod_time = (uint64_t) tv.tv_sec * 1000000 + tv.tv_usec;
+
+	return 0;
+out_remove_disk:
+	del_gendisk(gnbd_dev[i].disk);
+out_remove_file:
+	class_device_unregister(&gnbd_dev[i].class_dev);
+out_unregister_class:
+	while(i--){
+		del_gendisk(gnbd_dev[i].disk);
+		class_device_unregister(&gnbd_dev[i].class_dev);
+	}
+	i = MAX_GNBD;
+	class_unregister(&gnbd_class);
+out_unregister:
+	unregister_blkdev(major_nr, "gnbd");
+out:
+	while (i--) {
+		blk_cleanup_queue(gnbd_dev[i].disk->queue);
+		put_disk(gnbd_dev[i].disk);
+	}
+	return err;
+}
+
+static void __exit gnbd_cleanup(void)
+{
+	int i;
+
+	gnbd_ctl_cleanup();
+	for (i = 0; i < MAX_GNBD; i++) {
+		struct gendisk *disk = gnbd_dev[i].disk;
+		class_device_unregister(&gnbd_dev[i].class_dev);
+		if (disk) {
+			del_gendisk(disk);
+			blk_cleanup_queue(disk->queue);
+			put_disk(disk);
+		}
+		if (gnbd_dev[i].server_name)
+			kfree(gnbd_dev[i].server_name);
+	}
+	class_unregister(&gnbd_class);
+	devfs_remove("gnbd");
+	unregister_blkdev(major_nr, "gnbd");
+	printk(KERN_INFO "gnbd: unregistered device at major %d\n", major_nr);
+}
+
+module_init(gnbd_init);
+module_exit(gnbd_cleanup);
+
+MODULE_DESCRIPTION("Network Block Device");
+MODULE_LICENSE("GPL");
+
+#ifndef NDEBUG
+MODULE_PARM(debugflags, "i");
+MODULE_PARM_DESC(debugflags, "flags for controlling debug output");
+#endif
diff -urN linux-2.6.9/include/linux/gnbd.h linux-2.6.9-patched/include/linux/gnbd.h
--- linux-2.6.9/include/linux/gnbd.h	1969-12-31 18:00:00.000000000 -0600
+++ linux-2.6.9-patched/include/linux/gnbd.h	2004-10-22 13:39:34.000000000 -0500
@@ -0,0 +1,103 @@
+/******************************************************************************
+*******************************************************************************
+**
+**  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
+**  Copyright (C) 2004 Red Hat, Inc.  All rights reserved.
+**
+**  This copyrighted material is made available to anyone wishing to use,
+**  modify, copy, or redistribute it subject to the terms and conditions
+**  of the GNU General Public License v.2.
+**
+*******************************************************************************
+******************************************************************************/
+
+#ifndef LINUX_GNBD_H
+#define LINUX_GNBD_H
+
+#define GNBD_DO_IT	_IO( 0xab, 0x20 )
+#define GNBD_CLEAR_QUE	_IO( 0xab, 0x21 )
+#define GNBD_PRINT_DEBUG	_IO( 0xab, 0x22 )
+#define GNBD_DISCONNECT  _IO( 0xab, 0x23 )
+#define GNBD_PING	_IO( 0xab, 0x24 )
+#define GNBD_GET_TIME _IO( 0xab, 0x25 )
+
+enum {
+	GNBD_CMD_READ = 0,
+	GNBD_CMD_WRITE = 1,
+	GNBD_CMD_DISC = 2,
+	GNBD_CMD_PING = 3
+};
+
+#define gnbd_cmd(req) ((req)->cmd[0])
+#define MAX_GNBD 128
+
+/* values for flags field */
+#define GNBD_READ_ONLY 0x0001
+
+/* userspace doesn't need the gnbd_device structure */
+#ifdef __KERNEL__
+
+struct gnbd_device {
+	unsigned short int flags;
+	struct socket * sock;
+	struct file * file; 	/* If == NULL, device is not ready, yet	*/
+	int magic;
+	spinlock_t queue_lock;
+	spinlock_t open_lock;
+	struct list_head queue_head;/* Requests are added here...	*/
+	struct semaphore tx_lock;
+	struct gendisk *disk;
+	pid_t receiver_pid;
+	struct semaphore do_it_lock;
+	int open_count;
+	struct class_device class_dev;
+	unsigned short int server_port;
+	char *server_name;
+	char name[32];
+	unsigned long last_received;
+	struct block_device *bdev;
+};
+
+#endif /* __KERNEL__ */
+
+/* These are sent over the network in the request/reply magic fields */
+
+#define GNBD_REQUEST_MAGIC 0x37a07e00
+#define GNBD_REPLY_MAGIC 0x41f09370
+#define GNBD_KEEP_ALIVE_MAGIC 0x5B46D8C2
+/* Do *not* use magics: 0x12560953 0x96744668. */
+
+/*
+ * This is the packet used for communication between client and
+ * server. All data are in network byte order.
+ */
+struct gnbd_request {
+	uint32_t magic;
+	uint32_t type;	/* == READ || == WRITE 	why so long */
+	char handle[8];  /* why is this a char array instead of a u64 */
+	uint64_t from;
+	uint32_t len;
+}
+#ifdef __GNUC__
+	__attribute__ ((packed))
+#endif /* __GNUC__ */
+;
+
+/*
+ * This is the reply packet that gnbd-server sends back to the client after
+ * it has completed an I/O request (or an error occurs).
+ */
+#define SIZE_OF_REPLY 16
+struct gnbd_reply {
+	uint32_t magic;
+	uint32_t error;		/* 0 = ok, else error	*/
+	char handle[8];		/* handle you got from request	*/
+};
+
+struct do_it_req_s {
+        unsigned int minor;
+        int sock_fd;
+};
+typedef struct do_it_req_s do_it_req_t;
+
+#endif /* LINUX_GNBD_H */